Tracking deformed object and estimating motion parameters using point correspondence

There are many circular-shaped objects in our daily life. Most of these objects usually are deformed and move in both translational and rotational. Under prospective projection, a circle sometimes deformed into an ellipse. Some ellipse-based tracking algorithms can be used tracking the translation when the contour can be extracted automatically, but the rotation can not be extracted. In this paper, we propose an approach to tracking a deformed circular-shaped object with translation and rotation and estimating the motion parameters accurately. Our idea is to track some feature points on the object. Then we build the geometric relationship between the feature points and center of circle; Finally, we estimate the motion parameters by the geometric relation. We test the proposed approach using both simulated data and practical video of weight-lifting. For the simulated data, the average error for center of object is under 1.5 pixels, and the average error of rotation angle is under 0.50 degree. For the practical data, we can obtain the smoothing trajectory of center of disc. These experimental results show that the proposed approach is accurate and efficient.